Methodology

Data

• Window: blocks 52,392,731–52,695,131, 2026-10-09 19:40 → 2026-10-16 19:40 UTC (7.0 days). We chose a ~7-day sample because free RPCs cap log queries at 1,000 blocks.
• Events: USDC (0x8335…2913) AuthorizationUsed logs, the EIP-3009 event every x402 “exact” payment emits.
• x402 filter: we keep only transferWithAuthorization calls sent by one of 128 facilitator wallets listed in x402scan's open-source registry (Coinbase CDP, PayAI, Fluxa, Primer, thirdweb and others). Payer, payee and amount are decoded from the calldata.
• Facilitators seen: coinbase 438,693, openfacilitator 1,350, fluxa 1,347, primer 857, daydreams 95, openx402 77.
• Sources: tenderly and publicnode free public Base RPC endpoints. Scripts: fetch_base.py, fetch_funding.py, analyze_base.py, build_site.py.

Wallet-level screens

A payment counts as real only if it trips none of these flags:

• Self-payment: the payer and payee are the same wallet.
• Circular flow: the payee also paid the payer through x402 in the window (A↔B), or the payment is part of a 3-wallet cycle (A→B→C→A).
• Buyer funded by the merchant: outside x402, the merchant's wallet sent USDC directly to this buyer during the window. This check covers the top 150 merchants by volume.
• Tiny repeated payments: the same buyer paid the same merchant 100 or more times, averaging $0.01 or less.
• Single-buyer merchant: only one wallet paid this merchant in the window.
• Buyer concentration: one buyer wallet accounts for 90% or more of a merchant's USD volume in the window. Only that dominant buyer's payments to the merchant are flagged; smaller buyers are not.
• Exchange flow: the receiving wallet looks like an exchange deposit or withdrawal flow, not agent commerce. Exact rule: over the window, the receiver sent USDC directly to a wallet on our exchange wallet list (refunds to an exchange wallet that itself paid the receiver don't count), and those transfers total between half and twice what it received through x402. A payment to such a receiver is flagged if, within 30 minutes after it, the receiver sent at least half the payment's amount to a listed exchange wallet, or if the buyer was funded directly by a listed exchange wallet before paying (Base only). We use 30 minutes because exchange deposit sweeps we checked land within 3 to 6 minutes, and a longer window would start catching merchants' ordinary cash-outs. This check covers the top 150 Base merchants by volume.

We base these on TRM Labs and arXiv:2607.12575.

Limits

• Base and Solana are covered; other chains are not, so totals are smaller than x402scan's multi-chain 30-day figures.
• It only sees facilitators in the registry. Unlisted facilitators and non-USDC tokens are missed.
• “Same funder” clusters, where many buyers are funded by one outside wallet, are not screened yet. That needs each buyer's full funding history, which costs too many free RPC calls.
• Merchant funding uses direct USDC transfers only. ETH transfers and transfers routed through intermediaries are not checked.
• The exchange list is small and hand-checked: only wallets that Basescan or Solscan publicly tag as a centralized exchange (53 wallets, each with its source and check date). Individual users' deposit addresses are not listed; we spot them by their sweeps into listed wallets. Exchanges and wallets we don't list are missed.
• Passing the screens does not prove a payment was made by an AI agent.

Solana method

• How x402 settles on Solana: the buyer signs a transaction with one SPL Token transferChecked of USDC (EPjF…Dt1v); the facilitator pays the fee and submits it. Payer = transfer authority, payee = owner of the receiving token account.
• Filter: successful transactions whose fee payer is one of the Solana facilitator wallets in x402scan's open-source registry, with exactly one USDC transfer. Facilitator-paid transactions without a USDC transfer (e.g. account setup) are excluded.
• Coverage: 290,264 facilitator transactions in the window; 67,889 decoded as payments. Sampled: figment — 5,000 of 205,216 transactions decoded at random. Sampled payments are counted once (not scaled up), so Solana totals are a lower bound.
• Screens: the same seven as Base, including buyer concentration and exchange flow. Merchant-funding screen covers the top 50 Solana merchants by volume. Funding is detected from USDC the merchant wallet itself sent to its buyers outside x402. On Solana the exchange-flow screen checks only what those merchants sent to listed exchange wallets; buyers' funding from exchanges isn't fetched.
• Facilitators seen: payai 43,918, coinbase 16,694, dexter 4,037, figment 3,213, openfacilitator 23, openx402 2, ultravioletadao 2. Source: free public Solana RPC (or a free Helius key when configured).